Tissue transglutaminase drives fibrin β-chain cross-linking: a novel fibrin modification observed in trauma patients

Abstract

Covalent cross-linking of fibrin by the plasma transglutaminase coagulation factor XIII (FXIII) is a key determinant of blood clot stability and function. FXIII-catalyzed formation of ε-N-(γ-glutamyl)-lysyl cross-links is restricted to the fibrin γ- and α-chains and follows thrombin driven fibrin polymerization. Fibrinogen is also cross-linked by tissue transglutaminase (TG2) in a reaction favoring intra- and intermolecular α-γ cross-linking. Emerging evidence points to fibrinogen as a relevant substrate of TG2 in conditions of acute tissue damage. Remarkably, beyond detection of prototypical FXIII-directed cross-links (i.e., α-α, γ-γ), we identified entirely novel covalent cross-links involving the fibrinogen β chain (i.e., β-α, via FGB-Q82). Addition of TG2 to in vitro clotting reactions and analysis of fibrin(ogen) in reducing conditions revealed loss of β chain polypeptide paired with formation of high-molecular weight β chain species. Mass spectrometry-based cross-linking proteomic analysis of in vitro clots recapitulated the precise TG2-directed β chain cross-links observed in clots made using the plasma of trauma patients. The results document in vitro and ex vivo cross-linking of the fibrin β chain and highlight a novel example of TG2 emerging as a relevant plasma transglutaminase.

Graphical Abstract

Introduction

Fibrinogen circulates in plasma as a hexameric complex consisting of two Aα chains, two Bβ chains, and two γ chains (FGA, FGB, and FGG respectively).¹ In the first step of fibrin formation, the serine protease thrombin cleaves fibrinopeptides off the Aα and Bβ chains.¹ Cross-linking of the α and γ chains by the plasma transglutaminase coagulation factor XIII (FXIII) plays a crucial role in stabilizing blood clots, contributing to hemostasis and wound healing.² These cross-links are vital for providing the structural integrity necessary to prevent untimely clot dissolution.^2–4 Notably, recent discoveries have begun to reveal additional complexities in the fibrin cross-linking process, suggesting that other transglutaminases may also contribute to clot formation, particularly under certain pathological conditions.^5,6

Pioneering work by Laszlo Lorand, who isolated tissue transglutaminase (TG2) from red blood cells, documented that TG2 forms α-γ cross-links in fibrin.⁷ This contrasts the primary plasma transglutaminase FXIII, which predominantly creates α-α and γ-γ cross-links.² This discovery suggested the possibility that TG2 may play a role in fibrin(ogen) cross-linking within injured tissues, and perhaps play a role in conditions of major tissue damage. Emerging experimental evidence supports a role for TG2 in tissue fibrin(ogen) cross-linking, and elevated TG2 activation has been reported in RBC’s under physiological or pathological hypoxia⁸. However, biomarker-based evidence of a role for TG2-cross-linking of fibrin in patients is lacking. Accordingly, we sought evidence of TG2-directed fibrin cross-linking in trauma patients, a notion fully aligned with TG2 emerging as a relevant transglutaminase in conditions of acute tissue injury.

Study design

Trauma plasma samples were obtained from the randomized Control of Major Bleeding After Trauma (COMBAT) clinical trial (NCT01838863).⁹ Clots were generated from three trauma patients with normal fibrinolytic measurements from rapid thromboelastography. Controls were obtained from 9 healthy individuals (IRB 12–1614). After recalcification, thromboplastin was used for initiation. Clots were allowed to form over 30 min at 37°C. Clots were processed, and cross-linking-mass spectrometry (CL-MS) proteomic analysis was performed on chromatographic fractionated peptide digests (8 fractions each). Data was searched with MSFragger¹⁰ for protein identifications and pLink2.0¹¹ for cross-link identifications. R was used for data processing and Circos¹² was used for data visualization. For western blotting, in vitro fibrin clots were made using purified human fibrinogen (FIB1, Enzyme Research Laboratories [ERL]) or Peak1 fibrinogen (FXIII free) and human α-thrombin (ERL) with or without recombinant human TG2 (Zedira), and analyzed by western blotting or CL-MS^6,13,14. Fibrinogen chains were detected by western blotting using polyclonal anti-fibrinogen β chain antibody (Proteintech, Cat No. 16747–1-AP) or monoclonal anti-fibrin β chain (59D8, Sigma Aldrich, Cat No. MABS2155). Proteomic characterization was carried out as for the patient clots. Further details can be found in the supplemental methods. The Collection of Whole Blood or Blood Components for Use in Research (protocol #12–1614). Colorado Multiple Institutional Review Board (COMIRB protocol #: 12–1349) for study of trauma patient response to treatment.

Results and discussion

Clots were generated from three trauma patient plasma samples with normal fibrinolytic activity, as defined by thromboelastography (TEG). Cross-linking mass spectrometry analysis identified approximately 250 proteins in stringently washed clots, with nearly 75% of the signal corresponding to fibrinogen chains (FGA, FGB, and FGG). Within fibrin(ogen), key glutamine cross-link sites on the Aα chain that have been documented in previous structure–function studies^13,15,16 are shown in Fig. 1. Corresponding lysine sites on FGA were distributed throughout the αC connector to the C-terminal portion of the αC domain (19 total, from K225 to K625). Cross-linking within the FGG C-terminus was observed. For example, the FGG residues Q424/425 were found in cross-links with FGG K432 and eight distinct FGA lysine sites giving rise to previously reported γ-γ and γ-α cross-links (Fig. 1A and Sup. file 2.). Overall, the results validate the capacity of cross-linking mass spectrometry to discern precise cross-links in fibrin(ogen) formed during clotting reactions.

Figure 1. Chord diagram showing the cross-linking profile of fibrin clots.

A.) Clots formed from healthy controls. Starting at the top of the circle and moving clockwise FGG (red), FGB (orange) and FGA Q sites (yellow to green) found in cross-links. From the bottom clockwise to top are FGG then FGA, N- to C-terminal K sites (light blue to pink). The length of the outer arc (forming the circle) is based on the maximum number of cross-link identifications across the patient samples scaled to the maximum number across trauma patients and controls. The chord shows connectivity of Q sites (right half) to K sites (left half) and widths are based on the number of cross-links. The inner arc matches the color of the corresponding site of connectivity. B.) Clots formed from banked trauma plasma. C.) Structure of fibrinogen (PDB 3GHG) with the αC, β N-termini and γ-A C-termini regions modeled. Relative position of cross-linked lysines (K) domains (matching panels A and B) shown for the three chains on the left half of fibrinogen and glutamines (Q) on the right half.

To our surprise, we also identified several novel cross-links involving interconnection between fibrin Bβ and Aα chains in plasma clots from trauma patients. Fibrin Bβ chain, glutamine residue 82 (FGB Q82, shown in Fig 1C.) was cross-linked 14 times with three distinct FGA lysine residues (K437, K599, and K620) within the αC region of FGA (Fig. 1B, orange chords. The identified cross-linked peptides with assigned fragmentation spectra can be found in Sup. Fig. 1. FGB Q82 shown on a modeled region of the structure, Fig. 1C). These cross-links were conspicuously absent in clots generated from healthy control plasma (Fig. 1A), suggesting a trauma-specific mechanism. The αC region of fibrinogen is a substrate of FXIII-dependent reactions,^17,15 but to our knowledge, FGB is not a FXIII substrate. Thus, we hypothesized that the formation of FGB–FGA cross-links could be mediated by an alternative transglutaminase. Traumatic tissue injury or hemolysis may liberate intracellular enzymes such as tissue transglutaminase (TG2).⁶ To determine if TG2 drives fibrin β-chain cross-linking, we generated clots in vitro using pathologically relevant levels of TG2.⁶ Thrombin addition to purified human fibrinogen induced fibrinopeptide A and B removal and FXIII-dependent γ-γ dimer formation (Fig. 2A). In contrast, TG2 produced α-γ hybrid cross-links, even in the absence of thrombin (Fig 2A). Interestingly, while fibrinogen β-chain (~55kD) levels were stable in thrombin-clotted samples, the addition of TG2 reduced fibrinogen β-chain levels, even in reactions using FXIII-free fibrinogen (Fig. 2A–C), consistent with observations not elaborated on in prior publications.¹⁴ We posited that this was a consequence of TG2-dependent covalent incorporation of FGB into larger complexes. Indeed, a polyclonal FGB selective antibody detected numerous high molecular weight (HMW) bands in TG2-cross-linked fibrin (Fig. 2D). To assure these HMW bands were indeed β-chain, we used a monoclonal antibody (59D8) selective for a cryptic epitope in the fibrinogen β-chain¹⁸ revealed only after thrombin proteolysis (Fig. 2E). Importantly, the 59D8 monoclonal antibody detected HMW fibrin(ogen) β-chain in thrombin-clotted fibrinogen in the presence of TG2 (Fig. 2F). The results suggest that TG2 covalently incorporates fibrin(ogen) β-chain into HMW complexes in the presence or absence of thrombin.

Figure 2: TG2 cross-links the fibrinogen β chain to form high molecular weight complexes.

(A) Purified human fibrinogen (FIB1) and FXIII-free fibrinogen (Peak 1) were incubated with thrombin (i.e., to initiate fibrin polymer formation and FXIII-mediated cross-linking) or recombinant human tissue transglutaminase (TG2), as described previously.¹⁴ Reduced samples were resolved by SDS-PAGE and total protein visualized with Coomassie blue. For (B), FIB1 (2 mg/mL) was cross-linked by TG2 (25 μg/ml) and reduced samples were resolved by SDS-PAGE and visualized with SimplyBlue. A representative gel of at least 3 independent experiments is shown and fibrinogen β-chain quantified by densitometry (C). (D) Representative western blot for fibrinogen β-chain showing high molecular weight (HMW) cross-linked β-chain in fibrinogen cross-linked by TG2. (E) Thrombin (1 U/mL) or TG2 (25 μg/ml) was added to FIB1 (2 mg/mL). Representative western blots show fibrinogen β-chain detected by western blot using fibrinogen β-chain polyclonal antibody (top) or β-chain selective anti-fibrin (59D8) antibody (bottom). (F) Representative western blot showing high molecular weight (HMW) fibrinogen β-chain in fibrinogen cross-linked by TG2 detected using the anti-fibrin (58D8) antibody. For panels D-F, membranes were cut and probed with distinct primary antibody concentrations to resolve β-chain and HMW cross-linked β-chain.

We next generated a comprehensive map of the cross-linking profile of these clots using CL-MS. Thrombin-clotted (FXIII-cross-linked) fibrinogen displayed a cross-linking pattern dominated by FGA Q256 to αC-C K sites (109 identified across 3 samples) and FGG Q424/425 to K432 (99 IDs, represented by gray chords in Fig. 3A. Notably, FGB cross-links were conspicuously absent in clots made with only thrombin (and endogenous FXIII). The addition of TG2 dramatically altered the cross-linking profile. With the exception of a few FGA-FGA cross-link (Q347-K238, Q347-K625, for example), the addition of TG2 produced all the Q and K cross-link categories observed in trauma clots, including increased Q240/242 cross-linking in the N-termini of the αC region (αC-N, Fig 3.), FGG to FGA cross-linking (alongside a reduction in FGG-FGG), and most notably, FGB Q82 cross-links to multiple FGA αC sites (Figure 3B). TG2 also produced cross-links in all cross-link categories in reactions with FXIII-free fibrinogen, with a drastic shift toward FGG-FGA cross-links at the expense of FGA-FGA cross-linking (Figure 3C). Importantly, prototype FGB Q82 cross-links observed in trauma patients were produced by TG2 even in the absence of FXIII. Collectively, the results indicate that TG2 dramatically shifts fibrin(ogen) cross-linking patterns, including formation of unique FGB Q82 cross-links. These same cross-links were discovered in clots from trauma patients but not clots generated from healthy plasma. Indeed, TG2 plasma levels were elevated in the plasma of trauma patients (n=32) compared to healthy control plasma (1.3 ng/mL) and exhibited a trend to increase with the severity of trauma (NISS=1, 9.4 ± 9.5 ng/mL vs. NISS ≥ 10, 21.4 ± 23.2 ng/mL, P=0.09). The results are consistent with TG2 emerging as a relevant plasma transglutaminase in the context of trauma.

Figure 3. Chord diagram showing the cross-linking profile of in vitro fibrin clots formed with FXIIIA and TG2.

A.) cross-linking pattern of FIB with co-purified FXIIIA. Starting at the top of the circle and moving clockwise FGG (red), FGB (orange) and FGA Q sites (yellow to green) found in cross-links. From the bottom clockwise to top are FGG then FGA N- to C-terminal K sites (light blue to magenta). The length of the outer arc is kept constant and is based on the maximum number of cross-links observed at that site across conditions. The chord widths are based on the number of cross-links. B.) cross-linking pattern of FIB with co-purified FXIIIA and human TG2. C.) cross-linking pattern of FXIIIA free FIB (peak1) with human TG2.

The results provide the proof-of-concept that fibrin(ogen) is a relevant in vivo TG2 substrate and suggest that TG2 may be an underappreciated driver of fibrin(ogen) cross-linking in settings of acute tissue injury. The identification of β-chain cross-linking in our study adds a new dimension to these findings, indicating that TG2 can drive more diverse cross-linking patterns than previously recognized. These findings also raise the possibility that TG2 may partially compensate for the absence of FXIII in vivo, potentially playing a role in both traditional hemostasis and wound healing^19,20. Our experiments, including those using FXIII-free fibrinogen, demonstrate that TG2 alone can generate robust fibrin cross-links, many of which overlap with those formed by FXIII, while others are distinct. Notably, these include unique linkages involving FGB Q82, underscoring TG2’s potential to contribute to fibrin(ogen) stabilization independently of FXIII. The precise effect of these unique TG2-mediated β-chain cross-links on fibrin(ogen) effector function is unknown but may encompass effects on fibrin(ogen) hemostatic and inflammatory functions,²¹ both of which are pivotal during trauma-induced coagulopathy. Our studies inform on future site-directed mutagenesis and production of mutant fibrinogen proteins to uncover the effects of β-chain cross-links. Likewise, strategies used to generate mice expressing FXIII-resistant fibrinogen²² may inform the design of mice expressing fibrinogen that is uniquely resistant to cross-links uniquely imposed by TG2, such as those involving FGB Q82.

Collectively, we report the novel finding of fibrin β-chain cross-linking, driven not by the canonical plasma transglutaminase FXIII, but instead by TG2. Using cross linking-mass spectrometry and gel-electrophoresis mobility, we identified β-α cross-links in fibrin clots formed in vitro using purified fibrinogen and TG2. Remarkably, these precise cross-links were evident in clots generated ex vivo using plasma from trauma patients, strongly suggesting a genuine contribution of TG2 to fibrin(ogen) cross-linking in the context of trauma-induced coagulopathy. This transformational discovery paves the way for further investigations into the differential roles of transglutaminase isozymes in clot formation, and their contributions to the altered hemostatic balance observed in trauma and other disease settings.

Key Points.

1. Trauma patient plasma clots feature entirely novel fibrin β-chain cross-linking not evident in healthy controls

2. Fibrin β-chain cross-links evident in trauma patient plasma clots are recapitulated in vitro by tissue transglutaminase

Data sharing statement:

For original data, please contact kirk.hansen@cuanschutz.edu or luyendyk@msu.edu. Global proteomic results can be found in the online version of this article. Raw mass spectrometry data and result files are available via ProteomeXchange with identifier PXD062639 and PXD062688.